Action synopsis: pose selection and illustration
ACM SIGGRAPH 2005 Papers
A new pose-based representation for recognizing actions from multiple cameras
Computer Vision and Image Understanding
Single image summarization of 3D animation using depth images
Computer Animation and Virtual Worlds
A line based pose representation for human action recognition
Image Communication
| Hi-index | 0.00 |
A method is presented for automatically extracting key frames froman image sequence. The sequence is divided into clusters of frameswith similar appearance, and the most central frame in each clusterdefines a key frame. Clustering is done using an extension of thenormalized cut segmentation technique based on the inter-framesimilarities. The similarity between every pair of frames in thesequence is determined from the spatial image characteristics viaashape matching technique. Our algorithm is demonstratedsuccessfully extracting 20 key frames for a tennis player in actionover a 30 second (900 frame) video sequence.